personality types of DARPA PMs,
and help to create loyalty
left. Leaving aside, for the moment, the risk of collusion (which will be treated later in this paper), this means that, by self-selection, PMs cannot be risk-averse (Carleton 2010).
Third, term limits create loyalty. PMs must be willing to relocate for four years. Because early exits from DARPA are unusual, after they have joined DARPA an early exit would send a strong nega-tive signal to potential future employers. Becoming a PM thus has high entry and exit costs, which can be expected to foster loyalty (Hirschman 1970).
In sum, if collusion risk is contained, the four-year rule creates in PMs a cadre of risk-tolerant, loyal, technical experts with an ethos and high personal incentives to achieve substantial results in a short timeframe. It should be noted, however, that such techniques on their own will not automatically generate DARPA-like results or even DARPA-like PMs.26
Decision-making. DARPA’s basic decision-making architecture is flat, rapid, and, once programs are approved, provides even more autono-my than VC funds.
Programs must be approved by the office director and then by the DARPA director. Programs are subject to witheringly intense reviews attended by other ODs and multiple PMs. Often there is a further layer of informal scrutiny: that of the Office of the Secretary of De-fense and—formerly—that of the Director of DeDe-fense Research and Engineering (DoD).
While the DARPA director has the ultimate say, a near-consensus is often implicitly required for a program’s approval. This combina-tion of hierarchical and collective decision-making can be expected to cut down on “type II” errors—the approval of bad programs. Once programs are approved, however, PMs usually have substantial au-tonomy. Such autonomy can then be expected to cut down on “type I” errors—failing to invest in good opportunities.
Three factors play important, varying, and, at times, controversial roles in DARPA decision-making.
The first factor is “vision” (Carleton 2010). To be approved, a new program should aim at a radical innovation (i.e., an entirely new ca-pability). It must have a challenging-but-plausible path to a solution, a path that could not be traversed without DARPA’s involvement. In
some cases a new PM may already possess this vision—but often it results from a set of formal and informal processes for gathering and connecting ideas from DARPA’s networks (ibid.; Fuchs 2010).
The second factor is that of long-term strategic thrusts (Bonvillian and Van Atta 2011). These are not formally laid down but can acquire informal status as the result of a particularly strong long-term vision by a director combined with a bottom-up aggregation of a range of similar programs developing organic relationships to each other.
Once a thrust is present, it eases the approval of any subsequent program fitting within it. The most notable example was information technology, which encapsulates the risks and rewards of such thrusts.
Over several decades information technology produced, arguably, DARPA’s greatest successes. At the same time DARPA’s commitment to information technology may have inhibited the organization’s flex-ibility (when the thrust was eventually curtailed in the 2000s it cre-ated an outcry).
The third factor, and a point of particular controversy, is the im-portance of a “customer”: a military service or, on occasion, a civil-ian sector committed to implementing solutions once DARPA has brought them far enough along.
Once the services are convinced that they are capable of almost single-handedly implementing a new technology, they can and often will provide political cover for its development. PMs may find, within the same institutional umbrella, a “customer” able to both protect and implement the program.
Alternatively, this need for a “customer” sometimes becomes a screening device. In doing so it may provide discipline but may also block creativity, particularly if the program wishes to tackle a chal-lenge so difficult that its solution cannot yet even be outlined.
Adding such a test implicitly adds a level of hierarchy to program approval, shifting the balance of risk toward rejecting good programs and away from approving a bad ones. In striking that balance, direc-tors’ estimates of which risk is more serious probably influences the degree to which they impose such a test—in itself, this is perhaps a judgment of whether the agency’s culture at the time is thought to be too permissive or too conservative.27
These three factors, within the context of the structural conditions described above, produce a delicate balance of creativity and
disci-pline. The emphasis on vision enables the pursuit of radical innova-tion which alone can meet the demanding mission. Balancing this there is a risk that, as former German Chancellor Helmut Schmidt once said, “people who have visions should go see a doctor.”
Constraints seeming to keep DARPA’s visions healthy are the sim-plicity of the mission; requiring strategic thrusts to be validated by bottom-up successes in programs; the aggregation of “visions” from interactions and connections among DARPA’s network of the best minds in their fields; and, when it is used, the customer-demand test.
This delicate balance is perhaps best articulated in the “Heilmeier Catechism,” consisting of the following series of questions:28
• What are you trying to do? Articulate your objectives using abso-lutely no jargon.
• How is it done today, and what are the limits of current practice?
• What’s new in your approach and why do you think it will be suc-cessful?
• Who cares? If you’re successful, what difference will it make?
• What are the risks and the payoffs?
• How much will it cost? How long will it take?
• What are the “midterm exams” and “final exams” to check for suc-cess?
Every program must be able to answer these questions at approval and in reviews thereafter. The catechism connects DARPA’s mission to decision-making at the level of programs, using repeated interroga-tions to link the organization and progress of specific initiatives to the agency’s mission to generate technological breakthroughs.
Program Management. Once a program is approved, the PM’s task is to assemble, monitor, and manage research projects undertaking dif-ferent approaches to achieving a technological breakthrough. These are detailed in contracts specifying recipients, the approaches they will take, the milestones and targets for the research, and the funding that DARPA will provide.
During most of DARPA’s existence, PMs have retained a high degree of autonomy in writing programs’ contracts (the possible exception